How to Review 300 Contracts in Under an Hour
Traditional contract review requires attorneys to open, read, and analyze each document individually. At Mage, we built a workflow that lets you review hundreds of contracts systematically, without cutting corners on quality.
The Problem
A typical middle-market M&A deal involves 200 to 500 contracts in the data room. Reviewing each contract manually takes 15 to 30 minutes per document, meaning a single associate could spend 75 to 250 hours just reading through agreements. With deal timelines measured in weeks (not months), this math does not work.
300
Average contracts in a data room
20 min
Average manual review per contract
100 hrs
Total manual review time
The result: firms either staff deals with large teams of junior associates (driving up cost) or accept the risk that some contracts will only receive a cursory review. Neither outcome serves the client well.
The Workflow
Mage replaces the read-every-page approach with a structured extraction-and-review workflow. Instead of reading 300 contracts sequentially, you define what you need to find, let Mage extract it from every document simultaneously, and then review the results in a structured grid that highlights the items that require your attention.
Key insight: Most of the time in contract review is spent reading irrelevant boilerplate. By extracting only the provisions that matter, you eliminate the time spent on everything else.
Step-by-Step
Upload your data room
Drop the entire data room folder into Mage. Our parser handles PDFs, Word documents, and scanned images. Documents are classified by type automatically, so you do not need to pre-sort them.
Define your extraction scope
Select the provisions you want to extract: change of control, assignment restrictions, termination rights, indemnification caps, or any custom clause. You can use pre-built templates for common diligence workstreams or create your own.
Review the extraction matrix
Mage extracts the relevant provisions from every contract and presents them in a tabular grid. Each row is a contract; each column is a provision. You can sort, filter, and group by any attribute. Click any cell to see the source text with a link to the original document.
Focus on exceptions
The matrix view makes patterns immediately visible. Contracts with unusual terms, missing provisions, or high-risk language surface naturally. You spend your time analyzing the 10-20% of contracts that actually need attorney judgment, instead of reading the 80% that are standard.
Quality Assurance
Speed is meaningless if the results are unreliable. At Mage, we built multiple layers of quality assurance into the extraction workflow.
Source verification
Every extracted value links back to the exact page and paragraph in the source document. You can verify any finding with one click.
Confidence scoring
Each extraction is assigned a confidence score. Low-confidence results are flagged for manual review, so you know exactly where to focus.
Cross-document validation
When the same provision appears in multiple documents (e.g., a master agreement and an amendment), Mage cross-references them to surface conflicts.
Audit trail
Every extraction is logged with its source reference, extraction method, and confidence score. The audit trail gives you a defensible record of your review.
Frequently Asked Questions
Does this replace human review?
No. Mage handles the extraction and organization so that attorneys can focus on analysis and judgment. The attorney still makes every substantive decision. Mage just ensures you are looking at the right information, faster.
What about scanned or image-based PDFs?
Mage includes built-in OCR for scanned documents. Image-quality PDFs are converted to searchable text before extraction begins. The system flags documents that could not be processed (e.g., severely degraded scans) so you know which ones need manual attention.
How accurate are the extractions?
Our extraction pipeline achieves over 95% recall on standard provision types. For details on how we measure accuracy and minimize false positives, see our guide on Understanding False Positive Rates.
Can I customize the extraction criteria?
Yes. In addition to pre-built templates for common M&A workstreams, you can define custom extraction criteria using natural language descriptions of what you are looking for. The system supports both structured fields (dates, dollar amounts, party names) and unstructured provisions (qualitative clause language).